UHF partial discharge monitoring in GIS and signal processing

In this paper the experiments of detection partial discharges in GIS using ultra high frequency technique is described. The sensors are put outside the GIS so they do not bring extra requirements to the fabrication and operation of GIS. Some steps are taken to resist the external noises. The results show that UHF method is effective in detecting PD in GIS and avoiding the disturbance of coronas outside the GIS. Nevertheless very large corona, fluorescent tube or radio frequency noise nearby can still disturb the measurement. A method of signal processing by software is presented in this paper. Through the procedure of signal wavelet transform, signal noise extraction and signal reconstruction, the signal is analyzed to determine the magnitude of PD in GIS and avoid false alarm. And the modulus maxima and their positions gotten from the wavelet transform can be used as the base of PD pattern recognition in GIS.